The more precise you are about predictions, the more wrong you turn out to be. We will, therefore, start with two...
rather woolly predictions: that the days of the cottage industry local area network (Lan) and of cable spaghetti are nearly over.
It is also a pretty safe bet that we are in for a huge surge in bandwidth on the wide area network (Wan), while on the Lan it will continue to grow at its present rate. And there are no prizes for guessing that wireless communication will become far more prevalent within buildings, complementing long-distance mobility.
It is also likely, given recent extension of Gigabit Ethernet technology to the wide area, that the traditional division between the Lan and the Wan will cease to exist, other than as a natural physical boundary. All forms of data, including voice and video, will be carried as IP packets over a common physical infrastructure, even if there remains a logical separation. This means that instead of multiple cable types, there will be just one cable to each desk, and often none, with wireless links used instead, especially for connecting portable devices to the network.
These strands will all be interwoven to some extent, so that the trend towards integrated Lan/Wan services carrying everything will encourage users to outsource all networking from the desktop outwards.
This leads to the prediction from Andy Robinson, consultant at networking equipment supplier Equador, that the days of cottage industry Lans maintained in-house will be over, as it will make far more sense to hand the lot to a carrier such as BT, a systems integrator like IBM, or an outsourcing company such as EDS.
Similarly, the disproportionate increase in long-distance capacity, closing the gap on the Lan, will have an impact on the physical media within buildings. It will ensure that wireless does not take over completely, because only cable will be capable of carrying the volumes of traffic that the Wan will be able to deliver.
The improving price and performance of optical transmission, coupled with its potential for end-to-end delivery of high-speed broadband services, will encourage a large-scale migration from copper to fibre optic cabling.
Such a move has been long predicted but is now imminent, says Fred Engel, chief technology officer at network system supplier Concord Communications. Robinson agrees, and both cite the same cause: the proliferation of Dense Wave Division Multiplexing (DWDM) technology over all-optical networks, enabling high-speed services to be turned on and off like a tap. Whether users will be able to provide these services themselves, or have to request their carrier to do so, remains to be seen, but it could be that the enterprise networks of five years time will be "self service".
DWDM allows the capacity of a given optical fibre network to be multiplied in principle by 1,000 times or more by splitting the overall light signal into multiple wavelengths. Each wavelength then carries a single channel of communication, each with the potential to carry as much as the whole fibre did before. This also creates the potential for carriers to issue their customers with a group of wavelengths, each dedicated to a particular service.
This calls into question the need for complex protocols to support quality of service so that, for example, delay-sensitive real-time traffic like voice is given priority within single streams of IP traffic. Instead each enterprise would segregate their traffic into multiple channels, as done previously, but with the crucial difference that it will all be IP, making it much easier to integrate voice, video and data at the application level, where it matters.
But in many respects the most exciting developments will occur at lower bandwidths in the wireless arena. Here the running has been made by the IEEE 802.11 wireless Lan standard, which supports speeds of about 10 megabits per second (mbps), similar to the original Ethernet and plenty for most office applications. It is shortly to be extended to 50mbps.
More excitement has been created by the Bluetooth short-range radio communication standard (see p50). Its initial function is as a cable replacement radio link for connecting mobile devices to any sort of peripheral at distances up to 10m, and looks set to evolve into an indoor equivalent of the mobile phone network.